Epidemiologic research on lung damage caused by humidifier disinfectants

In April 2011 a tertiary hospital located in Seoul, Korea reported several cases of severe respiratory distress of unknown origin in young adults. To find the route of transmission, causative agent and patient risk factors of the outbreak, an investigation of the epidemic was initiated. A hospital based case-control study was conducted to indicate that humidifier detergent use was the cause of the outbreak. This information led the Ministry of Health and Welfare of Korea issued an order that humidifier detergents should be withdrawn from the market. Here, we describe the major events of planning, execution, and interpretation of the study, and discussions between researchers and public authorities following the decision to perform an epidemiologic study, chronologically.


INTRODUCTION
In late April 2011, six perinatal female patients were admitted to the respiratory intensive care unit of a general hospital (hereafter A-hospital) in Seoul, Korea for a lung disease with unknown cause [1,2]. The disease began with slight respiratory symptoms, and progressed to respiratory failure unresponsive to any treatment [1]. The medical staff requested an epidemiologic study on this unreported new disease from the Korea Centers for Disease Control and Prevention (KCDC). Officials of the National Epidemiology Center of the KCDC visited A-hospital to investigate. An "Emergency committee meeting for severe respiratory disease of perinatal mothers" was held on May 6, in which KCDC officials, KCDC committee members, and professors in the departments of pulmonology, infectious dis-ease, pathology, radiology, pediatrics, and preventive medicine in A-hospital participated, to discuss emergence of a novel disease, epidemic potential, and urgency of epidemiologic study.
In the emergency committee meeting, disease states, prognosis, and various test results of the six patients reported by Ahospital were reviewed, and epidemic potential, possible etiology, and future planning were discussed. Considering the unusual clinical, radiologic, and pathological findings, and lack of response to treatments for other lung diseases, it was determined to be a novel disease. On the other hand, similar findings were observed in lung tissues of adult males who lived with the patients, and it was reported that a lung disease with unknown cause has also been found in children since 2006. Various causes were suggested, including novel viral infection, yellow dust, drugs, and humidifier use.
On May 7, the KCDC requested an epidemiologic study of A-hospital, with a research team composed of clinical and preventive medicine staff in A-hospital, outside specialists, and KCDC officials (epidemiologic investigators). The purpose of the epidemiologic study was to objectively define the disease with unknown cause, investigate the extent of occurrence, and determine its etiology, in order to prevent involvement of more patients.
After three months of epidemiologic study [2] and cytotoxicity testing, the KCDC officially announced on August 31, 2011 that humidifier disinfectants (HDs) should be banned from use, and ordered removal of HDs from the market on November 11, based on animal inhalation toxicity test results.
During the six months from May to November 2011, the etiology of a new disease was revealed based on epidemiologic study results and biological evidence, including animal tests. This report will describe the epidemiologic study of diseases caused by HDs that the author performed as principal investigator (May to July 2011), including the chronology, characteristics, lessons, and limitations of the study.

CHRONOLOGICAL FLOW OF THE EPIDEMIOLOGIC STUDY OF LUNG DISEASE WITH UNKNOWN CAUSE
This section chronologically describes the major events of planning, execution, and interpretation of the study, and discussions between researchers and public authorities following the decision to perform an epidemiologic study, from May 7, 2011 (officially decided in late May) to August 31, 2011, with the official announcement that HD was the cause (Tables 1 and 2).
-May 7: Based on the emergency committee meeting on May 6, an epidemiologic study was planned. At that time, the target disease was severe pneumonia with unknown cause during pregnancy. -May 12: According to the diagnostic criteria, perinatal female patients in A-hospital and suspected patients in other hospitals were identified. Since similar diseases were identified in family members of patients, adult males were also in-cluded in the study. -May 13: Among nine patients in A-hospital, two were deceased, two underwent lung transplantation, and seven were perinatal females. Additional nine patients were identified in other hospitals. Disease onset was from February to early May, with no subsequent reports of new patients [1]. Pathologic findings indicated lung damage by inhalation, and potential common underlying diseases and autoimmune disorders were excluded. -May 15: Respiratory specimens from patients were tested to result in exclusion of viral infections from potential causes. -May 16: Based on clinical data including findings from chest radiographs, operational criteria (working definition) for the disease were established [1,2]. However, study expansion outside of A-hospital was unpalatable due to the Personal Data Protection Act of Korea, and approval from the institutional review board (IRB) of each medical institution. -May 17: The electronic medical records of A-hospital were searched with keywords such as adult females, ground glass opacities (GGO), and diffuse and disseminated, to select suspected patients. Most pathological findings in lung tissue specimens indicated fibrosis rather than inflammation; so that the disease was expected to be caused by inhalation of an unidentified substance. -May 18: Target disease of investigation were determined to be a new disease, but similar pediatric cases which were previously reported as acute interstitial pneumonitis in 2006 were found [3]. However, initial epidemiologic study was  Identification of epidemic and measurement of scale 7 May Investigation of whether this was an epidemic; Since this was a severe lung disease of perinatal females that was not observed previously, it was considered to fit the definition of an epidemic Establishment of working (operational) definition 8 May-15 May Revision and complementation of operational criteria; Patients who showed similar clinical patterns other than perinatal females were identified, and common characteristics were derived by reviewing their radiologic, pathological, and clinical findings; Lung damage that showed ground glass opacities in radiologic findings and no response to existing treatments for lung disease defined the operational criteria 16 May-2 Jun Identification of patients according to operational criteria: Chest CT images of patients identified through search of electronic medical record were re-inspected, and patient group with corresponding clinical findings was identified; A total 28 patients were identified after 2011 Identification of epidemic characteristics through descriptive epidemiologic study 7 May-2 Jun Various characteristics of identified patients group (28 patients in Asan Medical Center) were reviewed (Note ) 5/26/2011: IRB decided to waive review of study on patient group -Temporal characteristics: clear seasonality, mostly found from early spring to early summer -Spatial characteristics: nationwide, no regional specificity -Demographic characteristics: mostly female patients aged in their 20s and 30s, relatively fewer male patients Establishment of hypothesis for epidemic cause -It was highly likely to be a disease caused by inhalation of toxic substance -In particular, considering that bronchiolar area was mostly invaded, it was suspected that the size of corresponding particles should be extremely small Hypothesis testing through analytical epidemiologic study 2 Jun-29 Jun Including materials with respiratory toxicity, case-control study was performed on known risk factors for respiratory disease (Note ) 2/22/2011: IRB approval for the case-control study 28 patients were identified as patient group; control group included patients admitted for respiratory and allergic disease and patients admitted for delivery of babies Through survey with 18 patients in patient group and 121 patients in control group, humidifier disinfectants were found to be the epidemiologic cause Stage 2 Experimental study for biological probability  [2]. -June 24: Constructed questionnaire was finalized for the survey among controls. -June 28: A committee meeting on "unknown cause lung damage" was held. -June 29: The face-to-face survey for control group with respiratory disease and peripartum women was completed. -June 30: Viral test results identified no suspected cause. Patients with allergy disease were diced to be added for control group. -July 1: It was decided to add investigation of the components of HDs and animal testing to the epidemiologic study. -July 4: Houses of familial cluster cases were visited, to collect samples of their living environments. -July 7: Elevated odds ratio (OR) of HDs, humidifiers, and fungi were observed [2]. -July 8: The Korea Institute of Toxicology was identified as the animal inhalation toxicity testing agency for HDs. -July 11: Additional surveys which were conducted among additional control group were completed. Due to the low feasibility of expanding participants outside of A-hospital, the case-control study was decided to be completed. -July 12: Study results -that HDs were considered as the epidemiologic cause -were reported to the KCDC and Ministry of Health and Welfare. It was necessary to perform an animal inhalation toxicity test, and to identify components of HDs in detail. -July 13: It was decided to recruit additional local community control groups [4]. In addition, the necessity of supplementing biological evidence with animal test results was discussed. -July 18: The final results were reported to the Ministry of Health and Welfare, KCDC, and the advisory committee of the epidemiologic study including 18 cases and 121 controls. Crude and adjusted (for sex, age, and fungal exposure) OR of HDs were 47.3, 48.8 respectively (Table 3) [2]. It was decided to further perform pneumonocyte and animal inhalation toxicity testing for HDs. -July 20: It was decided to initiate a study on pediatric patients starting in August. As HDs were indicated for cause of the disease, in-depth interview and home visits were scheduled to begin July 21. -July 25: It was decided to add questions to assess dose-response relationships for HD to the existing questionnaire, and to conduct a survey on a local community control group [4]. It was also decided that the KCDC would perform all epidemiologic studies in children [3,5], and all cell and animal studies. For epidemiologic study in children, the team at A-hospital joined the research team, and used a similar study design [3,5].  [5].
-August 1: Additional questions asked during the in-depth interviews were added for local community control groups questionnaires [4]. To obtain biological evidence for causality, it was decided to perform cytotoxicity and animal testing with intratracheal instillation, with results to be reported in August. -August 8: In-depth and home visit study of cases were terminated, and a total of 15 patients were evaluated. As a preliminary step for animal inhalation toxicity testing (to determine particle sizes emitted in the air), a HD generation test was conducted.
-August 11: Progress of cytotoxicity testing was reported, the 50% lethal concentration was calculated, and pneumonocyte apoptosis was observed.
-August 18: Particle emission test results by the Korea Institute of Toxicology showed that HDs were able to invade even terminal bronchioles, and tests confirmed cytotoxicity of HDs.
In addition, animal testing with exposure to HD by intratracheal instillation resulted in weight loss and histopathologic inflammation, which confirmed toxicity in animals. -August 30: Results of epidemiologic study, cell experiments, and particle emission tests were discussed in the advisory committee meeting composed of experts in "lung damage with unknown cause" (attendees included members of the epidemiologic study advisory committee; members of the expert committee in infectious disease, environmental health, toxicity, and pulmonology; the contracted research team; and KCDC officials). -August 31: A press conference was held to announce that HDs were suspected as a risk factor for lung damage with unknown cause, and avoidance of their use and release was recommended until the causal relationship was finally confirmed. -September 2: A preparative meeting for study on children was held. -September 22: The research contract were made to perform a nationwide investigation on cases of severe lung disease with unknown cause in 2011. -November 10: The second expert epidemiologic study committee meeting was held. Pathologic findings of mice exposed to HD were identical to those in lung disease patients with unknown cause. -November 11: The Ministry of Health and Welfare announced that HDs were the cause of the lung damage with unknown cause, and ordered the removal of six types of disinfectants. -From 2012 to date: No additional case of lung damage in adults and children due to HD has been reported. In addition, the case-control study on interstitial pneumonia in children was performed in the same way as the present case-control study, and also revealed HDs as the cause [3,5].

ANALYSIS OF THE EPIDEMIOLOGIC STUDY ON HUMIDIFIER DISINFECTANTS
The present study is rare for having identified a new cause of lung damage due to HD. Several study characteristics, limitations, and lessons should be noted.
First, this was a multidisciplinary study of a disease epidemic reported by A-hospital, and performed by a collaboration between the hospital, a research team of the school of medicine (over 20 individuals from the departments of pulmonology, radiology, pathology, pediatrics, and preventive medicine), epidemiologic investigators of the KCDC epidemiology division, and over 20 experts from outside universities and institutes.
Second, the cause of a nationwide epidemic was only revealed through data from A-hospital. Because of the lack of clinical consensus for an epidemic of new disease, difficulty with medical data collection due to the Personal Data Protection Act of Korea, and the absence of a rapid institutional review process for a public study, the epidemiologic study was performed through a single agency.
Third, electronic medical records were actively utilized to determine if this was a new disease, to estimate the epidemic scale, and to define operational criteria for the disease. In the process of summarizing early reported clinical, radiologic, and pathological findings, the present epidemiologic study used common keywords, such as GGO, diffuse, and disseminated. The keywords were applied to electronic hospital medical records; potential patients were rapidly selected and evaluated by medical staff and radiologists to identify corresponding patients.
Fourth, after epidemiologic identification of HDs as a risk factor, experimental studies were rapidly performed to prove biological relevance. Since HDs are manufactured on a large scale, experimental evidence showing biological probability was required; epidemiologic evidence such as size of cross ratio and dose-response relationships was also required. For determination of exposure environment and disinfectant concentration in animal inhalation toxicity testing, toxicology, epidemiology, and clinical research teams established a feasible plan, which resulted in objective evidence of toxicity.
Fifth, an epidemiologic study should be performed using methods based on clinical knowledge. Thus, it is necessary for clinicians and epidemiologists to cooperate closely in all decisionmaking and research. A lack of communication might result in different data sets and even different conclusions. Clinicians and epidemiologists of A-hospital performed the study using data of A A-hospital alone (since a multi-agency study was impossible); however, this seemed to result in an uncomplicated and successful study.